A speech processing apparatus for mechanically recognizing speech uttered by a user is actively studied nowadays. The user's speech may be recognized in a local terminal or by a server over a network. In one example. Patent Literature 1 discloses a technique for recognizing the user's speech by both a terminal device and a server and for selectively employing a result of the recognition performed by the terminal device or a result of the recognition performed by the server.